Polyfunctionality and Inflectional Economy / 75
Total Page:16
File Type:pdf, Size:1020Kb
LiLT volume 11, issue 3 December 2014 Polyfunctionality and inflectional economy Gregory Stump One compelling kind of evidence for the autonomy of a language’s mor- phology is the incidence of inflectional polyfunctionality, the system- atic use of the same morphology to express distinct but related mor- phosyntactic content. Polyfunctionality is more complex than mere homophony. It can, in fact, arise in a number of ways: as an effect of rule invitation (wherein the same rule of exponence serves more than one function by interacting with other rules in more than one way), as an expression of morphosyntactic referral, as the effect of a rule of exponence realizing either a disjunction of property sets or a morphomic property set, or as the reflection of a morphosyntactic property set’s cross-categorial versatility. I distinguish these different sources of polyfunctionality in a formally precise way. It is inaccu- rate to see polyfunctionality as an ambiguating source of grammatical complexity; on the contrary, by enhancing the predictability of a lan- guage’s morphology, it may well enhance both the memorability of complex inflected forms and the ease with which they are processed. 1 Introduction In the domain of inflectional morphology, polyfunctionality is the use of the same morphology in the expression of distinct morphosyntactic property sets. Inflectional polyfunctionality is observable both within and across paradigms and even within individual word forms. It is not, in general, a marginal or exceptional phenomenon, but is often tightly integrated into an inflectional system’s structure. It is in some ways a paradoxical phenomenon: on one hand, it contributes to an inflectional system’s complexity insofar as it sometimes entails morphological am- biguity; on the other hand, it contributes to an inflectional system’s LiLT Volume 11, Issue 3, December 2014. Theoretical and Computational Morphology: New Trends and Synergies. Copyright © 2014, CSLI Publications. 73 74 / LiLT volume 11, issue 3 December 2014 economy insofar as it allows a small number of exponents to express a large number of morphosyntactic distinctions, and in some circum- stances heightens the predictability of unencountered word forms as well as the frequency of particular word forms or formatives (hence, presumably, their memorability or ease of processing).1 Inflectional polyfunctionality may arise from simple homophony: in such instances, two or more distinct inflectional exponents happen to be phonologically identical while differing in their morphosyntactic con- tent. For example, the suffix -s might be seen as exhibiting this trivial kind of polyfunctionality in the English sentence John like-s pear-s: the first instance of this suffix identifies the third-person singular present indicative form of the verb like, while the second instance identifies the plural form of the noun pear. In an inferential-realizational2 theory of morphology, this sort of polyfunctionality may be seen as the effect of two or more rules of exponence that introduce the same morphology for the expression of different content, as in Figure 1. Thus, suppose (a) that rules of exponence have the format in (1), where X is a metalinguistic variable over stems, C is a category of stems, τ is a morphosyntactic property set, and f is a morphological operation; (b) that a rule in this format is applicable to the pairing Z, σ of a h i stem Z with a morphosyntactic property set σ if and only if Z belongs to C and τ is a subset of σ; and (c) that the result of applying a rule in this format to Z, σ is the pairing f (Z), σ . In that case, the h i h i 1In writing this paper, I have benefited from discussions with Olivier Bonami and Berthold Crysmann; I also received very helpful suggestions from two anony- mous referees. I employ the following abbreviations: abl = ablative case obj = object rel = relativized argument dat = dative case pers = personal gender sbj = subject fut = future tense pl = plural sbj = subject inflection gnl = general tense prs = present tense sbjv = subjunctive mood ifd = Identity Function Default psm = possessum sg = singular ind = indicative mood psr = possessor neg = negative polarity pst = past tense 2An inferential theory of inflectional morphology is one that represents the re- lation between morphosyntactic property sets and their inflectional exponents as rules by which fully inflected word forms are inferrable from their stems; inferen- tial theories therefore differ from lexical (= lexicalist!) theories, in which relations 6 between morphosyntactic property sets and their exponents are stated in the lex- icon. A realizational theory of inflectional morphology is one that treats a word’s association with a particular morphosyntactic property set as a precondition for its inflectional realization; realizational theories therefore differ from incremental the- ories, in which a word acquires its morphosyntactic properties as a concomitant of acquiring its inflectional markings. See Stump (2001) for arguments for preferring inferential-realizational theories of inflectional morphology. Polyfunctionality and inflectional economy / 75 FIGURE 1Inflectional polyfunctionality arising from simple homophony trivialf polyfunctionality of English -s might be attributed to the pair of rules in (2). ; and (c) that tf (1) Rule of exponence: X, C, τ-sf (X) → (2) Two rules of exponencef in English (a) X, Noun, {pl} Xs → (b) X, Verb,s {3 sg prs ind} Xs s → The question naturally arises whether all polyfunctionality should simply be seen as involving simpleALL homophony of the sort exemplified in (2). The response that I propose here is: definitely not. In a carefully articulated theory of inflection, polyfunctionality may arise in several ways, and in many instances involves exponents that are not merely alike in form, but are in fact introduced by the very same rule of ex- ponence. Three kinds of evidence lead to this conclusion. First, there are phonologically identical inflectional exponents that, despite differ- ing in their content, are not merely alike in form, but participate in the same idiosyncratic allomorphy; treating these as distinct morpho- logical markings fails to account for their shared morpho(phono)logical idiosyncrasy. Second, there are phonologically identical inflectional ex- ponents that, despite their different content, nevertheless share a part of•their content; treating these as distinct morphological markings fails to account for their overlap in content. And third, there are phonologi- cally identical exponents that, though differing in content, nevertheless conform to a larger, systematic pattern of identity observable over a range of exponents; treating these as distinct markings that happen to coincide in form fails to accountpolyfunctionality for this larger across pattern. blocks • Given any polyfunctional exponent p, it is important to ask: where are p’s contrasting functions found? Fundamentally, there are three possibilities: . First, p’s contrasting functions may be associated with different 76 / LiLT volume 11, issue 3 December 2014 layers of words’ inflectional morphology. For example, the same affix may express distinct but related functions in more than one affix position. In inferential-realizational theories of inflection, it is customary to model each layer in a word’s inflectional marking by means of a dedicated block of realization rules. Thus, this first type of polyfunctionality might be formalized as a single rule of exponence serving distinct but related functions in more than one rule block. I shall refer to such instances as involving polyfunctionality across blocks. Second, p’s contrasting functions may be associated with different word forms realizing a given lexeme. This variety of polyfunction- ality might be modeled as a rule of exponence serving distinct but related functions in the realization of distinct cells in a lexeme’s paradigm. I shall therefore refer to these as instances of poly- functionality across cells. Finally, p’s contrasting functions may be associated with word forms realizing cells in the paradigms of lexemes belonging to distinct categories, including syntactic categories but also sub- categories of individual syntactic categories. I shall refer to such instances as involving polyfunctionality across categories. I consider cases of each of these sorts, for each of which I propose a different theoretical interpretation. In none of these cases should poly- functionality be equated with mere homophony, but must instead be seen as an intrinsic property of particular rules of exponence. In §2, I argue that instances of polyfunctionality across blocks involve a phe- nomenon of rule invitation wherein rules of referral realizing distinct but related content refer the realization of that content to the same rule of exponence in different rule blocks. In §3, I argue that instances of polyfunctionality across cells in- volve two distinct phenomena. One class of cases involves referral: in these cases, the exponence of one or more property sets in an inflec- tional paradigm patterns after that of some other property set by means of a rule of referral (§3.1). The complementary class of cases involves shared exponence: in these cases, two or more property sets have the same exponence, but there is no basis for regarding one of the prop- erty sets as providing this exponence as a model for that of the other sets; instead, a single rule of exponence causes two or more contrast- ing property sets to have the same